The AI Slop Wave: Why More Content Does Not Mean More Value

The recent surge in AI-generated e-books, which has tripled Amazon's monthly release volume, shows that content production has become disconnected from what readers actually need. While large language models (LLMs) have made it easy to mass-produce text, they have not created the breakout hits that defined past digital shifts. This trend creates a crowded middle ground that hides high-quality work, pushing legacy publishers to adopt proprietary, human-verified AI models to protect their brand. For readers and businesses, the competitive edge is no longer about producing more information, but about curating verifiable, opinionated, and high-quality signals in an increasingly noisy environment.

The Illusion of Productivity

The release of LLMs in 2022 triggered a massive increase in e-book supply. Researchers Imke Reimers and Joel Waldfogel found that monthly new e-book releases on Amazon grew from 100,000 to 300,000 by mid-2025, a trend that aligns with the adoption of AI tools.

However, this volume is misleading. In economic terms, quality is defined by market appeal, or what people actually buy and enjoy. By this measure, the AI-generated surge is failing. These titles consistently receive fewer ratings, lower star averages, and poorer sales rankings than books written by humans.

"AI does not seem to be like that. It is giving us more products but they are not products all the way across the distribution. It is a whole bunch more near the middle."

-- Joel Waldfogel

The market is currently flooded with mediocre content. Unlike the digital revolution of the mid-2000s, which produced cultural phenomena like 50 Shades of Grey from obscure fan fiction, the current AI boom has not produced a breakout hit. The system is scaling output, but it is failing to scale quality or cultural relevance.

The Credibility Gap in Expert Domains

The impact of this slop is most noticeable in sectors that require deep, verified expertise, such as travel writing. Jeremy Tarr, digital editorial director at Fodor's, notes that some authors have published 100 guidebooks in a very short time. This is physically impossible for human-led research.

When the cost of production drops to near zero, the market loses the signal of effort. In traditional publishing, the time and labor invested in a 500-page guidebook serve as a proxy for reliability. When that signal is removed, the system becomes prone to information rot.

"I think a lot of AI is fairly sycophantic and we wanted to make an AI that is opinionated. If you say should I go here or here they will not give you an honest assessment."

-- Jeremy Tarr

Fodor's is responding by building Eugene, a proprietary AI trained on their own verified content. By moving from a general-purpose model to a domain-specific, human-backed AI, they are attempting to reclaim the opinionated quality that generic LLMs lack. This is a defensive move: if they do not control the AI interface, they risk being relegated to raw data sources for competitors who will scrape their work to build inferior, sycophantic versions of their own expertise.

Why Human-Backed is the New Moat

The long-term consequence of the AI-slop wave is a flight to quality. As the volume of mediocre, AI-generated content grows, the value of human-verified, opinionated, and trustworthy information increases.

The strategy of human-backed AI acknowledges a critical system dynamic: AI is excellent at organizing and speeding up the retrieval of information, but it is limited by the quality of its training data and its tendency toward consensus-driven, sycophantic outputs. By injecting human editorial judgment into the AI loop, publishers like Fodor's are creating a product that is not just more information, but better guidance. The competitive advantage here is not in the volume of the output, but in the editorial friction: the refusal to be agreeable and the insistence on providing an honest, verified assessment.

Key Action Items

• Audit Your Information Sources: Over the next quarter, evaluate whether the tools you rely on for professional insights are aggregating slop or providing verified, human-backed expertise.
• Prioritize Opinionated Signals: Shift your consumption toward sources that demonstrate a clear, distinct point of view. In an age of AI-driven consensus, opinionated is a filter for quality.
• Invest in Proprietary Data: If you are a business, identify the unique, non-scraped data or expertise you own. This is your primary defense against commoditized AI competitors over the next 12 to 18 months.
• Shift from Volume to Verification: If you are a content producer, stop competing on output frequency. Focus on building human-in-the-loop systems where your editorial judgment is the primary differentiator.
• Build Friction into AI Adoption: When implementing AI tools for internal processes, ensure they are trained on your specific, high-quality data rather than general models. This prevents the sycophantic output problem and maintains brand integrity.